Webapp for exploring data from the COVID-19 chest xray image dataset
This webapp allows you to explore the images and associated metadata from the
COVID-19 chest xray image open
dataset. I built it as a
way of teaching myself fullstack web development using Vue
and Django. There is a live instance for people to
play around with.
Note: Upon accessing the live instance, you may see no images displayed for as
long as 15 seconds after page load. This has to do with limitations of Heroku’s
free tier.
The app displays a paginated gallery of all of the images contained in the
dataset. Clicking on an image will display a detail view where a larger version
of the selected image is displayed alongside detailed metadata for that image.
The app is setup to be deployed in two parts: the backend to heroku, and the
frontend to vercel. In the case of the frontend, the repo
currently contains no vercel-specific configuration, so in principle the
frontend could be deployed to any frontend deployment platform, provided that
the platform supports ignoring the python configuration (or simply detects the
node config first). This is currently an issue on netlify, for example.
While this app is primarily for demonstration purposes, there are hypothetically
a lot of ways in which it could be enhanced. The most useful unimplemented
feature would be a set of filters, allowing the user to selected a subset of
images that share one or more metadata fields. The queries could take advantage
of the client-side routing that’s already in place, allowing them to be
bookmarked and shared with others. If user accounts were added, this could
potentially evolve into a diagnostic tool, where users are able to upload their
own xray photos, and a ML model on the backend tries to provide a diagnosis of
the image. (However in principle this would perhaps be unwise, because as the
dataset README states, one should not implement a diagnostic ML model without a
clinical trial, and I am not a doctor.)